ANFIMAU | Healthcare monitoring system modernization

Overview

Healthcare Monitoring System establishes bi-directional communication with client medical cardiac devices, runs analytics on top of devices telemetry for tracking patient health changes, sends remote commands to cardiac devices, presents analysis results and alerts to doctors as well as provides integration with hospital systems.
As part of scaling to new markets / countries, extending the number of clients, establishing new partner integrations, new monetization models and enablement of cutting-edge AI-driven use-cases the customer faced challenges scaling its system in a cost-efficient way and with a low Time to Market for new capabilities.

Challenge

Monolithic Java-based platform architecture with centralized Oracle database, components tight coupling, hard storage limits lead to risks of enablement of new features, inability to support parallel development and handle continues data growth.
Inability to support AI-driven business use-cases due to lack of required platform capabilities, infrastructure constraints and data spread across variety of systems.
Lack of interoperability and data exchange to support new partner integrations via e-Health standards.
Costly and not scalable on-prem infrastructure prevents from scalability to new markets.

Approach

Synchronized replatforming & capability expansion

Parallel work streams for replatforming of the healthcare monitoring platform and new capabilities enablement to support new business initiatives.

Phased cloud-native transition strategy

Multi-staged replatforming of monolith healthcare monitoring system to cloud-native microservices approach, strangler & branch by abstraction patterns, event-driven microservice architecture style, Azure as a hosting cloud provider and A/B testing to support smooth transitioning.

Concurrent Azure ML platform development

In parallel development of ML Platform hosted in Azure to ingest / integrate datasets from existing systems, enable exploratory environments, feature store, model registry, continues training, model deployment, MLOps processes for AI-driven use-cases to support new business initiatives.

Data migration to Azure for AI-driven solutions

Data replication and migration from on-prem infrastructure to Azure based on progress of re-platforming and needs of new AI-driven or data exchange business use-cases.

FHIR integration for clinical data standardization

Usage of FHIR as a source of truth for clinical data and standard format for health information data exchange. Support of HL7v2 and DICOM standards and adapters into FHIR standard.

Efficient re-platforming & data migration execution

Planned re-platforming and data migration stages for the health monitoring platform were successfully completed that allowed it to easily add new features, scale infrastructure and business linearly within and outside of the current market in a cost effective way.

Cloud-based ML platform foundation establishment

Foundational ML Platform capabilities were created based on cloud technologies to support immediate AI-driven business use-cases.

Data Integration into Scalable Azure Data Lake

Data from multiple on-prem internal operational systems, legacy DWHs as well as from health monitoring platform was ingested and consolidated within scalable Data Lake hosted in Azure environment.

ML platform delivery and data lake enablement

Delivered ML platform, MLOps processes and data consolidation within cloud based Data Lake facilitated enablement of 2 new AI-driven business use-cases by client teams.

New partner integration via standardized data exchange

Enabled new partner integrations via FHIR, HL7v2 & DICOM standards for bi-directional data capturing and exchange.

Modern DevOps practices implementation

Implementation of modern DevOps practices like GitOps, Shift Left, Trunk-based development, Infrastructure-as-a-Code, Canary Deployment to increase time to market for new features with high quality and less risks to end clients.